i admit to confusing the strength of scavenge with reversion strength at idle. Burning my my daily candle from more than two points lately.
i admit to confusing the strength of scavenge with reversion strength at idle. Burning my my daily candle from more than two points lately.
2000 Ford Mustang - Top Sportsman
A stock cam at idle has virtually no overlap. So GM is not worried about the exhaust valve's role in this. We do worry about that with aftermarket cams, because we have lots of duration and overlap.
At idle, the engine is slow and there is no scavenging. With a big cam, we only have reversion. On a carbureted car, you can literally watch the fuel getting blown backwards and coming right out the top of the carburetor. Same happens with EFI, you just can't see it.
It's important to understand the injector duty cycle's role here. With some big injectors, the duty cycle at idle might be say 3%. Maybe less. There are 720 degrees of crank rotation per engine cycle, so 3% of that (a dataloggable number) is 21.6 degrees. So the injector is firing for 21.6 out of 720 degrees worth of possible time. If we wait until after the exhaust valve closes to START injecting the fuel, then the END of injection would be 21.6 degrees later than that. So let's say the cam closes the exhaust valve fully at 380 degrees. We need to END injection at 401.6 degrees or more, so the start of injection won't be until 380. Probably add a little more for an accuracy cushion in the real world. Basically at idle, we have all the time in the world to decide where to start and end injection. But don't forget we're referencing EOIT here, and have to calculate BOIT to make sure it starts after the exhaust valve closes.
In theory, we should do what GM does. Start injecting the fuel much earlier in the cycle, so that fuel can sit on that hot intake valve and vaporize. But if we do this too early, we might risk some other cylinders stealing this fuel with weird vacuum pulses in the intake coming from some other cylinder. So it's probably ideal to wait til later, and a hot valve won't take much time to vaporize the fuel. A cold valve won't be nearly as effective at that, and we should start injection much earlier, so there's more on-valve time.
In reality, our overlap causes the intake and exhaust valves to be open for a small period of time simultaneously. The exhaust is under either atmospheric pressure, or possibly even a small amount of backpressure. For us hot rod types, probably only atmospheric pressure since we have large exhaust. The intake valve, when open, is connected to intake manifold vacuum. So the pressure from the exhaust will push the air toward the vacuum from the intake when overlap occurs. This is how carburetor standoff happens. Anyway, that fuel that was sitting on that valve getting vaporized is now pushed backwards up into the common plenum, and ends up in some other cylinder's port that it wasn't intended for. Also, a bunch of inert exhaust gases get pushed into the intake. As I've said before, natural EGR.
Motors have a roughness to them during EGR flow. That's why the factory never opens the EGR valve at idle. And it's exactly the type of noises that cams make. Command some EGR flow at idle on a stock vehicle and notice that yourself. So the intake is full of fuel vapors that it shouldn't have in it, plus EGR gases that don't really run well, and carbon too. This leads to those other cylinders running a bit rough, and rich. As we improve the injector timing, we cure some of this richness, and that's why we can lean the engine out as and use air/fuel ratio as a bit of a guide to our effectiveness. Keep moving injector timing up if the motor keeps getting richer, allowing us to keep leaning it out. At some point, probably around 30 to 60 degrees of extra EOIT, the engine won't run rich any more and that means we've probably made all the progress we can.
At WOT, we SHOULD have about 80 to 85% injector duty cycle. But that depends on how well you picked your injector size for your engine's power level. But let's just say you datalog it and it is 80%. 80% of 720 degrees is 576 degrees. There is no way we can start injection at 380 and then inject for 576. That's way too late in the cycle. TDC is 0 on compression/power. BDC is 180 on power/exhaust. TDC again is on 360 exhaust/intake, BDC again is on 540 intake/compression, and 720/0 is compression/power again. The four accurate strokes of the motor are power, exhaust, intake, and compression because we start at 0 of the engine cycle, TDC compression.
By starting injection around 380 to degrees at idle in this example, we're done injecting by 401.6 degrees. The piston has just barely moved downward on the intake stroke. But in order to fit 576 degrees of injection into a 720 degree engine cycle, we're going to need to start that injection much earlier. We can get some idea of what's even possible by looking at the events in the last paragraph, along with a cam card. The intake valve isn't going to close until somewhere around the 560ish mark. We can't inject fuel before the intake valve closes, or it will get sucked into the last cycle, instead of the next cycle. So we're going to have to start injection at least after that. Let's say we start injection at 580. If we did, and injection lasted for 576 degrees, then 140 degrees left of this cycle, plus 436 degrees of the next cycle would put us at ending injection at 436.
Quite frankly, there just isn't that much movement we can make on WOT injection timing. In reality, maybe 60 to 80 degrees? The rest of the time just isn't available to us. If we can't start it earlier than 560 (intake valve closes), and we can't end it any later than that either, our EOIT has to fall somewhere between 436 and 560. This would change if our injector duty cycle were way too low, say 50% not 80%. And this is why oversized injectors might really benefit from playing with injector timing at WOT, compared to an appropriately sized or slightly small injector.
I haven't announced it yet because I haven't fully analyzed the data. But I did some serious testing of injector timing with my Picoscope, an in-cylinder pressure transducer, and measuring cam sensor, crank sensor, and injector pulse. Right off the bat, I'll tell you that a reference period is indeed 90 degrees. And there are 8 of them total, equaling the 720 degree engine cycle. The cam sensor on an LS1 is basically right at TDC compression, plus or minus a few degrees. It's only job is really to tell the EFI which stroke you're on, intake or exhaust. The actual injector firing events are timed off the crank sensor, which is high resolution.
I'll get into the details one day about how I tested repeated movements of injector delay, boundary, etc. Including setting them to 0 and seeing where they landed as I increased each one by 1.0 points. But the short story is that I largely came up with the same details as what BlueCat did, except the number -784 or whatever might not be exactly right. I think I came up with a number that aligned more with 760 or something like that. I'll post that in another post.
A stock LS1. The cam sensor would switch from high to low just a couple degrees before TDC. There was a little variance, I assume because of timing chain slack, etc at idle. Pretty freshly rebuilt motor though.
When I realized the cam sensor was within 10 degrees or less of actual TDC compression, I assume the engineers actually intended for the cam sensor transition to represent TDC, or close enough to it. So instead of trying to measure each of the following points off of real cylinder pressure, which is hard to do, I decided to take all my measurements based off of cam sensor transition instead, because that's after all what the EFI is going to base things off of anyway. I noticed that despite having all the best equipment hooked up to this thing that money can buy, there's still a bunch of slop that I can't account for. Each 1.0 of injector timing did not exactly equate to 90 degrees. It averaged it. I suspect the chain slack at idle just won't let perfection happen. At the end, I tried one WOT snap too, just to see about getting the chain slack out of it.
Here's the results. I started with leaving everything alone, and zeroing out normal. Most of these tests are with the warm engine idling. I added 1.0 normal to each test, flashed the car, and tested again back to back. I'm measuring the moment the injector pulse ends, or EOIT. And I'm rounding off to the nearest round number, not worrying about the decimals here.
Normal 0 = 543
Normal 1.0 = 617
Normal 2.0 = 704
Normal 3.0 = 74
Normal 4.0 = 166
Normal 5.0 = 261
Normal 5.55 (stock) = 309
Normal 6.0 = 351
Normal 6.5 = 396
Normal 7.5 = 485
Looking at the difference between each of these, I found them to be 74, 87, 90, 92, 95, 90, and 89. 74 was the 617-543. 90 was 351-261. 89 was 485-396. I skipped the 5.55 measurement because it wasn't exactly 1.0 apart from anything. other than the one 74 (which was my first measurement, and possibly not my best?), the rest closely followed 90 degrees apart, and almost certainly averaged 90. So mystery #1 is solved. One reference period is 90 degrees. It has nothing to do with crank teeth distance, number of crank teeth, GM referencing reference periods to crank, etc. That was a wild goose chase. It's 90 degrees, just like BlueCat said. And the obvious math of 8 x 90 = 720 just confirms what we already should have known.
Next, I returned Normal to stock, and tested Boundary in the same way, starting by zeroing it out...
0 Boundary = 443
1.0 Boundary = 532
2.0 Boundary = 622
3.0 Boundary = 710
4.0 Boundary = 78
6.0 Boundary = 261
Stock Boundary = 307
And the differences came in at 89, 90, 88, and 88. Again, Boundary periods are 90 degrees. And again when set to stock I came up with 307, pretty close to the previous stock test of 309.
If stock Boundary = 6.50 and Normal = 5.55 on a warm engine, and if the cam sensor is supposed to transition right at TDC compression (at least in theory, if not reality), then we should have 0 + (6.5 * 90) + (5.55 * 90) = 0 + 585 + 499.5 = 1,084.5 = 720 +364.5. We got more like 308. So where is the discrepancy? Well, perhaps the cam sensor really isn't at 0 TDC. Also, the injectors are no doubt actually fired by the crank sensor anyway. The cam sensor seemed to be about 19 degrees before TDC compression in reality.
I did a couple more random tests. 3.0 Boundary + 3.0 Normal netted me 483 EOIT.
5.0 Boundary + 5.0 Normal = 119 EOIT.
0 Boundary + 0 Normal at 2000 RPM = 699
0 Boundary + 0 Normal at 3000 RPM = 7 to 10
A WOT snap with Boundary and Normal set to 0 = 675 to 686.
It was frustrating to not be able to completely nail down these tests, when using this much high tech gear. It just wasn't perfectly scientific no matter what I tried. BlueCat's -784 formula suggests that he found the starting point for the injector equation to be 64 degrees before TDC. All the test numbers above that I gave you referenced the cam sensor transition, because that was easy to measure against. But my cam sensor looks to be about 19 BTDC itself. So the 364.5 number we calculated above would right off the bat be reduced by 19, netting us 345.5 degrees. I measured 308 or so relative to cam transition, so somewhere it's off. That 308 was relative to the cam transition too, suggesting that the formula might be 364.5 - 308 = 56.5 before TDC, or 720 + 56.5 if using BlueCat's strange negative number formula thing. That would be -776.5. He found -784. I would say that's within the margin of error, and certainly within the timing chain slop.
I originally thought his method might not be accurate due to piston dwell at TDC the way he was doing it on an engine stand. And that could still be. Unfortunately, I found that the running engine had a bunch of imperfect variables in it too... timing chain slop being the prime suspect. So I think that unless we do more testing on an engine stand, with a scope, and an 18" degree wheel dialed in to perfection levels with a solid adjustable pointer, I doubt we're going to do much better testing at this point. This is it.
I also don't think the exact numbers matter. We know that reference periods are all 90 degrees. And we know that adding -776.5 + 364.5 = 308, which is what I got. Why would GM start this formula out at 56.5 BTDC? I don't know. If BlueCat is right, it's 64 BTDC that the formula starts at. Heck we're both probably wrong and it's 60 even. I don't know. But I can tell you that the primary reason why might lie in ignition timing. Perhaps these signals have to be about 60 degrees earlier than TDC because the most timing you'll ever possibly need to command is 50 or so degrees, and we need to have a little time to calculate things on the fly on top of that. So 56 - 64 degrees before TDC for these signals to start seems feasible.
The crank sensor's 24 teeth are not all even in shape. Some are narrow teeth, some are wide, and the pattern is unique. The spot where actual TDC occurs according to cylinder pressure is on the rising edge of the first fat tooth, when you see the pattern of 5 skinny teeth and then 5 fat teeth that follow. So the 5 skinny teeth are at the end of a 360 degree crank turn, and the 5 fat teeth start at the beginning of a 360 degree crank turn. I'll try to spell them out here in order:
Fat, fat, fat, fat, fat, skinny, fat, fat, fat, skinny, skinny, fat, fat, skinny, skinny, skinny, fat, skinny, fat, skinny, skinny, skinny, skinny, skinny.
pontisteve, thanks for your efforts to take another look at Bluecat's formual. It's good to know that you have validated the 90 degree reference periods. As far as the 784 value, if I remember correctly from Bluecat's description of his work he questioned the value himself, indicated that his measurements could be off some, but not enough for him to feel that the real number was actually one that we might expect. I'm good with our formula being good +/- a few degrees.
I'd like to readdress something in my first post which I guess you missed regarding why stock EOIT is around 300-308 degrees, or at least state it in a different way. Assuming the stock boundary value represents the earliest that the engineers wanted the injector to fire at max stock duty cycle, which with the stock boundary value and the formulas you and Bluecat have published puts it at 225-233 degrees ATDC on the intake stroke, which sounds reasonable as that is around where stock the intake valve would just be closed, and then further assuming that the expected max stock injector duty cycle was 69-70%, you'd end the injector firing right where you measured it.
One other thing, you've stated many times that the injector needs to wait to fire until the intake valve has closed on the previous stroke. While from a purist point of view that seems prudent I don't think it is a big deal. Besides Greg Banish having stated so himself earlier in this thread, before upgrading to larger injectors (a second time) on my supercharged engine my data logs were showing duty cycles greater than 100%. Obviously I don't believe they actually were over 100% but I do believe the computer was calculating an injector firing time and passing that value over the interface even though it couldn't actually fire them that long for each event. I didn't notice any particular issues running the injectors that hard while making up for the limited fueling with water/meth injection.
Once again thanks for your efforts in helping us all figure this out by your measurements and talking through it.
Last edited by Hoosier Tuner; 02-15-2019 at 09:58 PM.
I'm not sure where you're getting your numbers from, but that's not exactly what I said. First off, the stock injector duty cycle that most OEMs are going to shoot for is likely to be 80 to 85%. However, I can't say I've tested that a bunch either. It's just what the math suggests, when looking at the injector size used in most cars vs the power they have to support.Originally Posted by Hoosier Tuner
I'd like to readdress something in my first post which I guess you missed regarding why stock EOIT is around 300-308 degrees, or at least state it in a different way. Assuming the stock boundary value represents the earliest that the engineers wanted the injector to fire at max stock duty cycle, which with the stock boundary value and the formulas you and Bluecat have published puts it at 225-233 degrees ATDC on the intake stroke, which sounds reasonable as that is around where stock the intake valve would just be closed, and then further assuming that the expected max stock injector duty cycle was 69-70%, you'd end the injector firing right where you measured it.
One other thing, you've stated many times that the injector needs to wait to fire until the intake valve has closed on the previous stroke. While from a purist point of view that seems prudent I don't think it is a big deal. Besides Greg Banish having stated so himself earlier in this thread, before upgrading to larger injectors (a second time) on my supercharged engine my data logs were showing duty cycles greater than 100%. Obviously I don't believe they actually were over 100% but I do believe the computer was calculating an injector firing time and passing that value over the interface even though it couldn't actually fire them that long for each event. I didn't notice any particular issues running the injectors that hard while making up for the limited fueling with water/meth injection.
As for injector duty cycles exceeding 100%, I don't know why GM's computer would show that. Except to say that it's calculating what would need to happen, now what is possible. Obviously, you can't turn an injector on more than 100% of the time. If an injector is on for so long that there isn't enough time to shut it off and then re-open it again, the PCM will just leave the injector on. This can actually lead to a rich condition in the 95 to 100% duty cycle range, followed by a lean condition if the motor needs any more fuel than what 100% supplies.
I don't know what you're referencing with that 225-233 number. If GM set up the Boundary number as the earliest an injector can fire, then that's SOIT not EOIT that we would have to measure. That doesn't make much sense exactly. Although we do know that they clearly started at 0 TDC and added Boundary, then added Normal to that. Notice above that my measurements (relative to the cam sensor transition near TDC, which may be a minor imperfection) shows the 0 point at approximately 675 to 10 (730). Setting Normal to 0 left the Boundary only, to determine the earliest EOIT point at 543. Now graph out the engine cycle over time. 0 = TDC compression, 180 = BDC power, 360 = TDC overlap, 540 = BDC intake, and repeat. So GM has chosen the EOIT boundary to be approximately right at BDC on the intake stroke. The intake stroke occurs from 360 to 540.
I'm not sure of the stock LS1 cam specs, but I bet the intake valve doesn't stay open too much longer than BDC, and they must not have wanted to keep injecting as the piston moved upward on the compression stroke.
As for the injector needing to close after the intake has closed on the previous stroke, it's because we want the fuel to go into the cylinder on the next event, not the last event. I mean, are we firing the injector this time for this cycle, or the next one? In practice, the engine may not care that much. Look at how a batch fire car runs. All injectors squirting all the time, or at least for one bank. None of them timed at all really. But that doesn't mean it won't run better with sequential injection. It will.
Regarding your first quoted paragraph above, I don't know how I got that either! Re-doing the math using Bluecat's equation puts GM's boundary around 160 degrees ATDC in the intake stroke. Apologies for the resulting confusion. (Using info from your post (#543) would seem to put it closer to 170 degrees.)
From what I've found online for stock LS1 cams, the 0.050 intake valve closing position is 42 degrees ABDC on the compression stroke, 81 degrees ABDC for the 0.004 closing position. Assuming for a moment the 60 degree BTDC on the exhaust stroke Bluecat's equation gives for EOIT, gives 399 degrees of rotation between the 0.004 closing point, the 0.050 point 438 degrees. Relative to the 720 degree cylinder cycle these angles relate to 55% and 61% duty cycles respectively. Using the value you found above of -776.5 instead of Bluecat's -784 would add 1% to these duty cycle calculations. Unless GM sized the injectors for sub 60% duty cycle they clearly weren't concerned about firing the injectors before the intake valve closing, at least not for higher rpm WOT conditions.
One final note, looking at a stock tune for a 2000 corvette that I pulled out of the tune repository, Power Enrichment adds 25% at 6000 rpm and above. That means 20% of the fueling at 6000 rpm is what Power Enrichment is adding. Maybe GM was ok with the Power Enrichment portion of the injector cycle being fired for the previous intake cycle???
I still don't know where you're getting these numbers from. We're measuring End of Injection here, not beginning. At WOT, the EOIT is somewhere near BDC on the intake stroke.
GM sprays fuel on the back side of a closed intake valve at light loads. This helps vaporize the liquid fuel. At WOT, there isn't enough time to spray onto a closed valve only, so they spray when the intake is both open and closed. But they don't continue to spray after the intake valve closes. At WOT, they spray before the intake valve opens, and while it's open. But not after it shuts, until the next engine cycle.
Bluecat found the stock tune parameters gave an EOIT at about 300 degrees ATDC of the start of power stroke, which indeed is on the back of a closed intake valve at light loads. Are you saying that EOIT doesn't stay the same? I believe EOIT is at fixed position, defined by the parameters in the tune, and the start of the injection is determined by how long the computer calculates the injector needs to be on and then subtracted from the EOIT.
Found this old post onLS1 tech: https://ls1tech.com/forums/pcm-diagn...l#&gid=1&pid=2
post#3
CKP_CMP_waveform_relationship_001.jpg
Adding the injector #1 trace would help.
Also in post#19: (thanks joecar)
if you look closely at the waveform you will see that:
- in each 360? there are 4 quadrants;
- in each 90? quadrant, there are 6 segments that uniquely encode that quadrant;
- in each 15? segment, there are 2 pulses, 3? and 12?, in either order, with opposite sense to each other.
The start of injection definitely gets back-calculated from EOIT. But now that you mention it, in all those tests I did, I was looking to see how Boundary and Normal affected EOIT, and proved that out pretty well. 90 degrees and all. But I never really tried to drive the car or test to see if EOIT changed when the tune did not.
Fords are so much simpler. A 3D table based on load and RPM gives you EOIT.
I decided to check mine against this, just to see if my cam and crank line up the same way this capture from Autonerdz had it. It did. The trailing edge of the 4th narrow crank pulse lines up perfectly with the trailing edge of the cam pulse transition from hi to lo.
And for fun, I did a little paint work to show the crank sensor pattern off better. The PCM detects piston position by seeing where the engine is in it's rotation. It's easier for it to determine it's position if it has many unique characteristics it can read. The PCM just counts skinnies and fats, and quickly knows where in the 360 degree crank cycle it's at. It looks at cam sensor signal to determine if it's on the intake stroke or the power stroke.
There are 24 crank pulses per crank rotation, 48 crank pulses per engine cycle. The transition from the end of one pulse to the end of the next pulse equals 15 degrees of crankshaft rotation. Piston movement is measured in crank rotation, so the PCM can tell where that piston is at, within 15 degrees, at all times. It only needs to count about 5 of these pulses before it knows where in this rotation it is, because that's the most number of times any pattern repeats itself exactly.
At 15 degrees each, the engine can figure out where it's at within 75 degrees of crank rotation. Notice all the crank pulses end at the same spot. But the skinny pulses begin later.
The blue signal at the top is the cam sensor. It's a 1x signal, meaning it's either high, or low. The PCM uses this to know which stroke the engine is on, compression or exhaust.
The yellow trace shows the injector firing event on a stock tune. The green trace is cylinder pressure. I added a black line to show you how the trailing edge of the cam sensor lines up with the trailing edge of the 4th skinny crank pulse in a row.
This illustrates my previous comments about how the cam sensor didn't exactly line up with TDC compression. On one side, you could say there may be a slight delay in the cylinder pressure sensor, and that in reality they do line up perfectly. You could also say this might be a cam or crank minor timing variation, and it really doesn't line up right. Probably a variation in where the crank sensor was pressed onto the crank.
If you've ever done a Crank Relearn, and read that it was used to determine the crank pattern so that misfire monitoring would be accurate, this is what the PCM is seeing. You put it in crank relearn mode, floor it, and when it hits the rev limiter, you let off the gas. The decelerating crank takes all the slack out of the timing chain and the PCM learns where each of these crank teeth are. Later, it will use that data when it detects a misfire, so that it can determine at what crankshaft degree range the misfire occurred. Then it will calculate which cylinder was supposed to be firing during this period, which is just basic math if you know the firing order. It can then throw a P030x code saying which cylinder was the cause of the misfire. It's imperfect, but decent at determining the offending cylinder.
Last edited by pontisteve; 02-19-2019 at 03:27 AM.
pontisteve, I hope you don't mind but I've taken the liberty of overlaying an 80% duty cycle on your last chart, assuming that the EOIT stays the same. As can be seen, this injector pulse starts in the heart of an intake stroke and continues pushing fuel until well after the intake valve closes. Most of the fuel goes into the following intake stroke.
Added 80% duty cycle.png
In post#553, should the duty cycle bar go in the opposite direction - EOI to the right?
If End Of Injection Time is fixed and the start time is then calculated to start a given time before that, then the duty cycle bar needs to go to the left.
Do the injection events occur as marked?
picockp.JPG
The chart is data measured by pontisteve.
ANyone have the latest version of the calculator that was created here? I couldve sworn I provided my own version here but hell if I can find it..
JRS SOIT EOIT Worksheet 041715.xlsEOIT.xlsxI have two versions.
Last edited by JimMueller; 04-21-2019 at 01:50 PM.
Very nice.
Now if i could find the math for load based.
Reply With Quote